Saturation transfer and chemical exchange measurements of the stereochemical drift occurring during the Wittig reaction

Article abstract of DOI:10.1002/mrc.1573

The Wittig reaction of butylidenetriphenylphosphorane with benzaldehyde using LiHMDS as base in THF was studied. The stereochemical drift (different ratio obtained in alkenes versus oxaphosphetane intermediates) was followed by low-temperature 1D NMR tech

Full text of DOI:10.1002/mrc.1573

MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2005; 43: 451–456
Published online 5 April 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mrc.1573
Saturation transfer and chemical exchange
measurements of the stereochemical drift
occurring during the Wittig reaction
Aurelia Pascariu,1 Mircea Mracec2 and Stefan Berger1∗
1
Institute of Analytical Chemistry, University of Leipzig, Johannisallee 29, D-4103 Leipzig, Germany
Romanian Academy, Institute of Chemistry Timisoara, B-dv Mihai Viteazul 24, 300223 Timisoara, Romania
2
Received 8 December 2004; Revised 2 February 2005; Accepted 7 February 2005
The Wittig reaction of butylidenetriphenylphosphorane with benzaldehyde using LiHMDS as base in
THF was studied. The stereochemical drift (different ratio obtained in alkenes versus oxaphosphetane
intermediates) was followed by low-temperature 1D NMR techniques. A retro-Wittig reaction is demon-
strated using 13C and 31P saturation transfer experiments and homonuclear DPFGSE-ROE techniques.
Copyright  2005 John Wiley & Sons, Ltd.
KEYWORDS: NMR; 1H NMR; 13C NMR; 31P NMR; Wittig reaction; saturation transfer experiment; DPFGSE-ROE
systems affects the stereoselectivity of the Wittig reaction
considerably.9
INTRODUCTION
The Wittig reaction1 is one of the most important trans-
formations in organic chemistry for the preparation of
carbon–carbon double bonds, being used in natural product
synthesis2 and in industrial processes.3 The double bond is
introduced in place of the carbonyl functionality, therefore
this reaction is regioselective and the stereochemistry can be
controlled. The reaction of unstabilized ylides and aldehydes
yields mainly (Z)-alkenes.1
Since we have been applying recent NMR methods
for the investigation of the reaction mechanism of the
Wittig reaction, we became interested in studying the
stereochemical drift10 occurring in the Wittig reaction
of butylidenetriphenylphosphorane (1) with benzaldehyde
as depicted in Scheme 1. The term stereochemical drift
was coined by Maryanoff and co-workers, who observed
different E : Z ratios in the oxaphosphetane intermediates
2 and 3 versus the finally isolated alkenes 4 and 5. The
difference was dependent on concentration, lithium base,
solvent, temperature and substitution. Chemical crossover
experiments, carried out by Maryanoff’s group, showed
that addition of a second and different aldehyde after
the complete formation of the oxaphosphetanes leads also
to alkenes containing the organic residue of the second
aldehyde, and this was interpreted as a consequence of a
retro Wittig reaction.
We could confirm these stereochemical findings and
observed in our hands a Z : E ratio of 71 : 29 for the
oxaphosphetanes 2 and 3, whereas for the alkenes 4 and 5 a
Z : E ratio of 52 : 48 was found. This is a 19% larger production
of (E)-alkene compared with (E)-oxaphosphetane. In addi-
tion to a retro-Wittig reaction, also an opening, rotating and
reclosing of the oxaphosphetane ring system either between
the former ylide carbon and phosphorus, between the former
aldehyde carbon and oxygen or between the former ylide
carbon and the former aldehyde carbon (with the rotation of
one bond) could explain the results (see Scheme 2).
Several mechanisms for the Wittig reaction have been
proposed and have attracted the interest of different research
groups. In our previous studies on the mechanism of the
Wittig reaction, it has been shown that using rapid injection
NMR (RI-NMR) techniques, at very low temperatures,4
a
new dynamic equilibrium between oxaphosphetanes and
their lithium salt adducts could be detected. After the
observation of the failure of the Wittig reaction in the
case of dipyridyl ketone in the presence of Li ions,5 we
could show the existence of stable betaine–lithium salt
adducts6 using solid-state NMR. This was also the case
when instead of dipyridyl ketone normal benzaldehyde
was used as the carbonyl compound, but one, two or
three phenyl rings in ethylidenetriphenylphosphorane were
replaced with pyridyl rings.7 Replacing the phenyl rings
with 2-furyl substituents using BuLi as base, we were able
to isolate a stable oxaphosphetane when all three phenyl
rings of ethylidenetriphenylphosphorane were replaced.8 In
general, the replacement of phenyl rings by heterocyclic ring
We were therefore interested in whether, by more
advanced NMR techniques, we could directly detect a
retro-Wittig reaction or an equilibrium between the two
stereoisomeric oxaphosphetanes without the use of chemical
crossover experiments and thus shed some more light to the
ŁCorrespondence to: Stefan Berger, Institute of Analytical
Chemistry, University of Leipzig, Johannisallee 29, D-4103 Leipzig,
Germany. E-mail: stberger@rz.uni-leipzig.de
Contract/grant sponsor: Fifth Framework Programme;
Contract/grant number: HPTM 2000-00183.
Copyright  2005 John Wiley & Sons, Ltd.

452
A. Pascariu, M. Mracec and S. Berger
Scheme 1. Mechanism of the Wittig reaction.
(a)
1 M
0.5 M
0.1 M
Scheme 2. Possible intermediates of the Wittig reaction.
-58 -59 -60 -61 -62 -63 -64
ppm
(b)
0.1 M - Z isomer
0.1 M - E isomer
0.5 M - Z isomer
0.5 M - E isomer
1 M - Z isomer
1 M - E isomer
A
B
C
D
E
F
90
80
70
60
50
40
30
20
10
A
Figure 1. Pulse scheme for the saturation transfer experiment
with three different r.f. channels, shown for the example of 31P.
200 220 240 260 280 300
Temperature K
mechanistic questions involved. As a central method within
this study we used the saturation transfer method (Fig. 1),
where dynamic equilibria are tested in the slow exchange
limit by selective irradiation of one component.11 Irradiation
of one signal leads to an intensity decrease of those signals at
all different chemical sites with which the irradiated signal
is in chemical exchange.
Figure 2. (a) 31P NMR for the Wittig reaction at different
reaction concentrations in THF at 203K; (b) stereochemical drift
according to the NMR integration of the oxaphosphetanes.
with the lithium ions. The resolution of the spectra decreases
with increasing concentration of the reactants in THF, also
owing to the large amounts of salt and phosphine oxide
formed. As a compromise, we chose a 0.5 M reaction concen-
tration for further investigation. At this point we still have a
reasonable resolution and a large stereochemical drift.
From Fig. 2(b), it can be seen that the stereochemical
drift occurs at different reaction concentrations, which is in
accordance with the results of Maryanoff et al.12 The graphic
[Fig. 2(b)] shows the difference in the E-isomer production
starting from the 203 K ratio of oxaphosphetanes, where the
RESULTS AND DISCUSSION
Figure 2(a) shows the low-temperature 31P NMR spectra at
203 K in the oxaphosphetane region at different concen-
trations. In the spectra it can be seen that at 0.1 M, the
oxaphosphetane signal appears as a sharp singlet; at 0.5 M
the (Z)-oxaphosphetane signal becomes broader and at 1 M
both oxaphosphetane signals are broad. From previous stud-
ies, we know that this broadening is caused by aggregation
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 451–456

Stereochemical drift during the Wittig reaction 453
first point results from 31P NMR spectra recorded directly
after adding the benzaldehyde, and the last one at 298 K
from the ratio of alkenes (from H NMR spectra) when the
reaction was finished. All intermediate points represent the
ratio of oxaphosphetanes determined by 31P NMR at the
given temperatures.
in addition spectra with irradiation in a region where no
signal was present as a negative control. The 31P saturation
transfer experiment was recorded using only 0.7 equiv. of
benzaldehyde to ensure that excess of ylide was present in
the reaction. In this respect, we irradiated the 31P signals
of the ylide, (Z)-oxaphosphetane and (E)-oxaphosphetane
and we recorded a spectrum with irradiation of the TPPO
formed in the reaction as a negative control, since it cannot
be assumed that the TPPO contributes to the equilibrium.
The difficulty with the saturation transfer experiment as
applied here is the circumstance that all signals decrease
owing to the ongoing chemical reaction. Hence each irradia-
tion experiment will invariably lead to a decrease in all other
signals, simply owing to the time that the irradiation exper-
iment needs. The experiments were therefore performed
with multiple times and at different temperatures. Owing to
the limitation in the exact reproducibility of organometallic
sample preparation, the signal intensities of the individual
experiments cannot be analyzed all together but, neverthe-
less, by careful comparison and statistical data analysis of
the individual decay curves we were able to detect exchange
between the irradiated species. To our knowledge, this is
the first time that the saturation transfer technique has been
applied during an ongoing reaction.
1
If one follows a Wittig reaction by means of NMR, one
1
has the choice of observation by 31P, 13C or H NMR. For
31P the signals of the starting ylide, the oxaphosphetane
signals and the signal of the final product triphenylphos-
phane oxide (TPPO) are of relevance. For 13C and 1H,
the signals of the starting aldehyde, the oxaphosphetanes
and the alkenes are of interest. If one labels the alde-
hyde carbonyl atom, an observation by 13C NMR is espe-
cially easy. Since no considerable line broadening was
observed for the 13C NMR signals, we chose to apply the
saturation transfer technique, which is known to extend
the limits of dynamic NMR in the direction of slower
exchange. For our 13C saturation transfer experiment we
used 98% labeled C6H513CHO. The ratio of starting materials,
C6H513CHO : butylidenetriphenylphosphorane, was 1.2 : 1.
We used excess of benzaldehyde to have the benzaldehyde
signal always present for the saturation transfer measure-
ments. Spectra were mainly recorded at 253 K where the
reaction is very slow but the stereochemical drift is already
occurring. We irradiated the 13C signals in benzaldehyde, (Z)-
oxaphosphetane and (E)-oxaphosphetane and we recorded
Results of 31P saturation transfer
After using standard 1D processing and careful integration
of the signals, the results in Fig. 3(a) and (b) were obtained.
(a)
(b)
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
PO irradiation
PO irradiated
Ylide irradiation
E-oxaphosphetane irradiation
Ylide irradiation
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
Z-oxaphosphetane irradiation
0
20 40 60 80 100 120 140 160 180
Time (min)
0
20
40
60
80 100 120 140 160 180
Time (min)
Figure 3. Time behaviour of the 31P NMR signals of the (a) (Z)- and (b) (E)-oxaphosphetane signals during the Wittig reaction at
253 K. The curves show the situation when the phosphane oxide signal was irradiated as a control ( ) and the situation when the
ꢀ
ylide signal ( ) or the other stereoisomer of the oxaphosphetane (ꢁ) was irradiated.
ž
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 451–456

454
A. Pascariu, M. Mracec and S. Berger
The curves fitted through the squares serve as controls
obtained by irradiation of the TPPO signal, which cannot
contribute to the equilibrium, hence these curves depict only
the decrease due to the chemical reaction. It is obvious that
the curves fitted through the circles and triangles, which
were obtained by irradiation of the ylide signal and by
irradiation of the other stereoisomer, are always lower than
the control curves, indicating a saturation transfer between
both the ylide and the oxaphosphetane and between the
oxaphosphetanes themselves. This decrease proved to be
statistically significant using the 3ꢀ criterion.
This may be due to the smaller amount present in the reaction
in comparison with the (Z)-oxaphosphetane.
Results of DPFGSE-ROE
To confirm these results further, we tried to use an
independent NMR method. Since 2D methods such as
2D 31P-EXSY proved not to be sensitive enough, we
turned to the 1H-DPFGSE-ROE technique,13 which, at
these temperatures, is not prone to sign change problems
compared with NOE. A mixing time of 1.4 s was shown
to give the best results.1H-DPFGSE-ROE gives higher
values for intramolecular interaction, but the intermolecular
interaction with the benzaldehyde can also be detected. As
a control, we also measured spectra in which the irradiation
was performed in a region without signals. Some typical
results are shown in Fig. 5, showing weak signals of both
the aldehyde proton at 10.1 ppm and the ortho aromatic
protons of benzaldehyde at 8.1 ppm on irradiation of the
oxaphosphetane signals.
The results of two independent NMR techniques on
three different nuclei all show that in the case of the chemical
system studied there is an ongoing equilibrium between
aldehyde and oxaphosphetanes as determined by 13C NMR,
between ylide and oxaphosphetanes as determined by 31P
NMR and 1H NMR. Hence the data confirm the double
arrows in Scheme 1. However, these results cannot, at
present, exclude one of the possibilities depicted in Scheme 2.
Results of 13C saturation transfer
Figure 4 shows the behaviour of (Z)-oxaphosphetane. The
curves fitted through the squares serve as control by
irradiation in a region far away from all 13C NMR signals.
Circles and triangles were obtained by irradiation of
the benzaldehyde signal and by irradiation of the other
stereoisomer. In the case of the benzaldehyde, the curve
obtained is always lower than the control curve, indicating
again saturation transfer between the benzaldehyde and the
(Z)-oxaphosphetane. The (E)-oxaphosphetane curve (data
not shown) is hardly distinguishable from the control curve.
Control curve
E-oxaphosphetane irradiation
Benzaldehyde irradiation
12
10
8
EXPERIMENTAL
Saturation transfer experiment. Spectral details
All spectra were recorded with a triple-channel Bruker DRX
400MHz spectrometer using a TBI probe head with an addi-
tional 13C coil or a BBO probe head tuned to phosphorus. All
31P chemical shifts were referenced using the  scale.14 The
saturation transfer experiments15 were performed according
to Fig. 1 with the triple-channel possibility of the instrument.
Typical data for the 31P measurements were a time domain
of 32K data points, a spectral width of 130 ppm with an
acquisition time of 0.78 s. The irradiation time d1 was chosen
6
°
to be 5 s and the pulse p1 was 5 µs, corresponding to 30
excitation. Eight scans were accumulated after four dummy
scans. The power of the 300 W amplifier for single-frequency
decoupling was attenuated by 69 dB. The 31P decoupling
bandwidth under these conditions was checked to be of the
order of 50 Hz, which was chosen since the oxaphosphetane
signals are typically separated by 300 Hz at 253 K.
The 13C saturation transfer experiments were performed
with the same pulse sequence, but using a spectral width
of 230 ppm and a time domain of 64K data points yielding
an acquisition time of 1.4 s. For 13C a saturation time of 10 s
was used and 16 scans were accumulated after four dummy
scans at 253 K.
For the integration of the individual spectra, one common
integral range file was created to ensure the same integration
areas over the whole time of the reaction. Automatic baseline
correction was used before integration and the integration
was performed with and without phasing of the individual
integrals.
4
2
0
50
100
150
200
250
Time (min)
Figure 4. Time behaviour of the 13C NMR signals of the
(Z)-oxaphosphetane signals during the Wittig reaction at
253 K. The curves show the situation when the spectra was
irradiated as a control (at a position where no signal was
present) ( ) and the situation when the benzaldehyde signal
(ꢁ) or the (E)-oxaphosphetane ( ) was irradiated.
ꢀ
ž
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 451–456

Stereochemical drift during the Wittig reaction 455
(a)
ppm
5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 4.6 4.5
(b)
(c)
10.5 10.0 9.5 9.0 8.5 8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5
ppm
Figure 5. (a) 1H NMR spectra, 0.5 M THF-d8, 253 K; the oxaphosphetane protons were irradiated as shown; in
(b) (E)-oxaphosphetane and (c) (Z)-oxaphosphetane. DPFGSE-ROE spectra were recorded at 253 K, 64 scans, mixing time 1.4 s, lb
2 Hz.
CONCLUSIONS
DPFGSE-NOE and DPFGSE-ROE. Spectral details
The 1D selective ROE spectra were obtained with the pulse
sequence as described in Ref. 13; 32 scans were accumulated
after four dummy scans with a time domain of 32K data
points, using a spectral width of 14 ppm, yielding an
acquisition time of 2.8 s and using a relaxation delay of
We have been able to prove, using NMR techniques,
results which had previously been shown only by chem-
ical crossover experiments. By saturation transfer experi-
ments at low temperature, we could demonstrate exchange
between reagents and reaction intermediates. Also, we con-
firmed these results with 1H-DPFGSE-ROE. Further, we shall
investigate direct exchange between oxaphosphetane inter-
mediates.
°
1 s. The selective 180 Gaussian-shaped pulse was 25 ms and
the gradient pulses used were in the ratio 70 : 30. The hard
°
90 excitation pulse was 10.3 µs. The power level for the
°
spin lock was set corresponding to a 21 µs 90 pulse. At this
°
power level, 30 pulses (7 µs) were used, followed by a delay
Acknowledgements
Financial support by the Fifth Framework Programme, as a Marie
Curie Fellowship, Contract HPTM 2000-00183, is acknowledged.
of 14 µs. The total duration of the spin-lock used was varied
between 300 ms and 2 s.
General procedure for Wittig reaction
REFERENCES
Phosphonium salt (0.5 mmol) together with solid LiHMDS
(0.5 mmol) was suspended under nitrogen in 1 ml of THF-
d8. The mixture was stirred for 30 min at room temperature
and a reddish orange solution was obtained. With careful
exclusion of moisture and air the solution was transferred
1. (a) Wittig G, Geissler G. Liebigs Ann. Chem. 1953; 580:
44; (b) Wittig G, Schollkopf U. Chem. Ber. 1954; 87: 1318;
(c) Maryanoff B, Reitz A. Chem. Rev. 1989; 89: 863.
2. (a) Nicolau K, Ha¨rter M, Gunzer J, Nadin A. Liebigs Ann./Recl.
1997; 7: 1283; (b) Zarbin P, Princival JL, de Lima E, dos Santos A,
Ambrogio B, de Oliveira A. Tetrahedron Lett. 2004; 45: 239.
3. (a) Pommer H. Angew. Chem. 1977; 89: 437; Angew. Chem. Int. Ed.
Engl. 1977; 16: 423; (b) Pommer H, Thieme C. Top. Curr. Chem.
1983; 109: 165.
°
to an NMR tube. At ꢀ80 C, 1.1 equiv. of benzaldehyde was
added directly to the NMR tube through a septum. When
measuring ylide behaviour, only 0.7 equiv. of benzaldehyde
was added.
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 451–456

456
A. Pascariu, M. Mracec and S. Berger
11. Mann BE. J. Magn. Reson. 1976; 21: 17.
4. Geletneky C, Fo¨rsterling FH, Bock W, Berger S. Chem. Ber. 1993;
126: 2397.
12. Maryanoff B, Reitz A, Mutter M, Inners R, Almond H Jr,
Whittle R, Olofson R. J. Am. Chem. Soc. 1986; 108: 7664.
13. Bauer W, Soi A, Hirsch A. Magn. Reson. Chem. 2000; 38: 500.
14. Harris RK, Becker ED, Cabral de Menezes SM, Goodfellow R,
Granger P. Pure Appl. Chem. 2001; 73: 1795.
5. Subramanyam C. Tetrahedron Lett. 1995; 36: 9249.
6. Neumann R, Berger S. Eur. J. Org. Chem. 1998; 1085.
7. Schro¨der U, Berger S. Eur. J. Org. Chem. 2000; 2601.
8. Appel M, Blaurock S, Berger S. Eur. J. Org. Chem. 2002; 1143.
9. Ackermann M, Pascariu A, Berger S. to be published.
10. Reitz A, Nortey S, Jordan A Jr, Mutter MS, Maryanoff B. J. Org.
Chem. 1986; 51: 3302.
15. Berger S, Braun S. 200 and More NMR Experiments. A Practical
Course. Wiley-VCH: Weinheim, 2004; 152.
Copyright  2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 451–456